Understanding Subtypes of Bullying Victims Using Target Congruence and Lifestyle-Routine Activities Theories: A Latent Class Analysis

Abstract

Prior research on bullying victimization has often explored the behavior from a binary perspective in which victims and non-victims are examined. However, there has been an increasing amount of literature taking a more nuanced approach towards different subtypes of victims in recent years. This study sought to examine heterogeneous patterns of victimization and identify correlates associated with different victim groups through an integrated framework of both lifestyle-routine activities and target congruence theories. Using a national sample of 2217 Korean adolescents, the latent class analysis uncovered three groups of victims including physical/verbal, verbal-only, and low/non-victims. The results supported the relevance of integrated target congruence and lifestyle-routine activities theories to explain subtypes of bullying-victims. Among the dimensions of target congruence, vulnerability and antagonism factors were distinctively associated with the three subgroups of victims. Bullying perpetration was relevant in distinguishing physical/verbal and verbal-only victims from non-victims even when target congruence variables were controlled for.

Keywords

Bullying victimization lifestyle-routine activities target congruence bullying subtypes latent class analysis

Introduction

Bullying victimization is distinguished from other forms of victimization in that it involves repeated aggressive behaviors stemming from a power imbalance between the victim and the offender (e.g., being overpowered by the perpetrator) (Olweus, 1993, 1994). Despite the definition widely used, scholars have found that juveniles who have engaged in bullying are more likely victims of bullying because, following the logic of victimization theories (i.e., delinquency is viewed as risky behavior that put a person in close proximity and exposure to potential offenders, leading to youth victimization) (Cho, 2017a; Park & Cho, 2021; Widom, 1989). Also, prior studies have found that youth with psychological distress, anxiety, physical limitations, impulsivity, and aggressiveness are more likely to be victimized by peers (Augustine et al., 2002; Frías & Finkelhor, 2017; Kahle & Peguero, 2017; Kulig et al., 2017; Zavala & Whitney, 2019).

Relative to research on individual characteristics, risky lifestyles, and its impact on victimization, less attention has been given to what theoretical perspectives facilitate such relationships. Furthermore, there is an increasing interest in the multi-faceted dimensions of bullying victimization and how various subtypes of bullying victimization may co-occur (e.g., Berkowitz et al., 2015; Bradshaw et al., 2013, 2015). Victims of bullying are assumed to consist of various distinct subgroups, each being subjected to different patterns of bullying (e.g., physical, verbal, social/relational, and cyber) that might be related to unique covariates and consequences (Choi et al., 2019). The Center of Disease Control and Prevention (CDC) acknowledges that bullying could take different forms and inflict verbal, physical, psychological, and/or social harm on victims (Gladden et al., 2014). Nevertheless, most research examining these topics have not accounted for the heterogeneity of youths’ bullying victimization, employing instead a variable-oriented approach in which relations among variables become the primary focus, thus being vulnerable to causal over-simplification.

Considering that some youth may be victimized by multiple types of bullying while others may be victimized by only one type, the current study examined the multi-faceted dimensions of bullying victimization. In doing so, this study employed a person-centered latent class analysis (LCA) approach to assess potential heterogeneity in bullying victimization among South Korean youth and better understand the co-occurrence of various types of bullying victimization. In a latent class analysis model, a latent categorical variable was created to determine Korean youth’s class membership, and their relationships with predictors such as victims’ characteristics and risky lifestyles drawn from the integrated approach of target congruence and lifestyles-routine activities theory. Prior research has conducted latent class analysis to examine patterns of bullying victimization among US adolescents (Nylund, Asparouhov, et al., 2007; Wang et al., 2010). However, relatively little is known about patterns/subtypes of bullying victimization among Korean youth and their associations of individuals’ class membership with victims’ characteristics—target vulnerability, gratifiability, and antagonism—as well as risky lifestyles. Therefore, the study aims to address empirical and methodological limitations of the extant research that used a single composite variable of bullying victimization on the application of latent class analysis and test the utility of the integrated approach of both target congruence theory and lifestyles-routine activities theory to South Korean youths. In addition to offering a cross-cultural applicability of theories of victimization, this study provides practical implications by identifying characteristics and risky lifestyles that makes Korean youths more vulnerable to multiple types of bullying victimization.

Review of Literature

Classifying and Assessing Subtypes of Bullying-Victims

Subtypes of bullying-victims

In 2011, the CDC reported that 18% of youths aged 12 to 18 in the United States (U.S.) were verbally bullied; 8% were physically bullied; and 5% were subjected to physical threats by another adolescent (Gladden et al., 2014). Physical bullying refers to perpetrators’ use of physical force, such as hitting, kicking, or pushing the victim. In contrast, verbal bullying refers to the infliction of harm through oral communication, including taunting, name calling, or insulting another individual. In recent years, research on bullying victimization has progressed from taking a binary approach (i.e., victim vs. non-victim) towards identifying and studying different forms and subgroups of victims. In addition to the distinction between victims, bullies, and bully-victims (e.g., Haynie et al., 2001; Holt & Espelage, 2007; Olweus, 2010), researchers have examined whether there are meaningful differences across varying categories of victimization. These studies have found that youths experience many different types of bullying victimization, with different forms displaying distinct levels of prevalence and empirical associations with other variables. Below, we review the utility of distinguishing victims of physical and verbal bullying. We also discuss how a more robust classification of bullying can be conducted through latent class analysis.

Latent Class Approach to Classification of Bullying-Victims

Existing research on subtypes of bullying indicate that substantial amount of victims report two or more types of verbal, physical, and relational victimization (e.g., Volk et al., 2006; Waasdorp & Bradshaw, 2015). Whereas these attempts relied on cutoffs determined by researchers, the current study categorized the sample into subgroups on the basis of common patterns extracted by statistical analysis. Latent class approach involves classifying observed discrete measures into a number of mutually exclusive latent classes (McCutcheon, 1987). Through an iterative process, individuals with heterogeneous victimization experiences are categorized into a number of subgroups containing similar patterns that best reflect the original data. This person-centered approach is considered more rigorous than relying on arbitrary cut-points because various statistical standards (i.e., fit indices, parameter estimates, standard errors) are used to identify the number of classes. Most of existing research employing latent class or growth mixture approach has focused on identifying patterns of bullying perpetration and victimization, finding a group of aggressive bullies, low-to-moderate bullies, moderate victims, non-victims, as well as a subgroup that reports both perpetration and victimization (e.g., Ashrafi et al., 2020; Ferreira-Junior et al., 2021; Liu et al., 2021; Martínez-Monteagudo et al., 2020). While it is helpful to distinguish victims and bullies, these studies do not tell us much about varying types and potential overlap in victimization (Lasky et al., 2021). It was our interest to employ this method in order to empirically categorize youths into a common pattern of subcategories of bullying victimization and assess potential overlap in the forms of victimization (Molenaar & Campbell, 2009; Nylund, Bellmore, et al., 2007).

Research using the latent class approach on discrete subtypes of bullying-victimization has identified three to four classes (Berkowitz et al., 2015; Bradshaw et al., 2013, 2015). These groups commonly include a no/low victimization group and a group associated with a high level of involvement in all forms included in the study. Across these studies, the no/low victim group accounted for the largest class (i.e., 50–85% of the sample), while the multiple victimization class typically comprised the least amount of the sample (5–11%). The remaining groups consisted of a combination of moderate to high level involvement in verbal, physical, and/or relational forms. For example, Bradshaw and colleagues (2013) detected four groups (low/normative, high verbal, physical, & relational, verbal & relational, verbal & relational) among middle school youths and three classes (e.g., low/normative, high verbal, physical, & relational, verbal) among high school youths. Similar patterns were found in Berkowitz and colleagues’ (2015) sample of middle and high school youths in California, where youths were classified into four groups consisting of no victimization, victims of frequent verbal and other forms, victims of occasional verbal and other forms, and verbal and sexual victims.

Also commonly found in these studies was a concentration of developmental problems among the multiple-type victimization groups. Regardless of school-level (i.e., middle school, high school), students who had a high probability of endorsing all forms of bullying victimization manifested significantly greater amounts of problems. They were distinguished from other groups, especially from the low/normative group, by higher scores of internalizing symptoms, aggression, and negative perceptions about school climate (Bradshaw et al., 2013, 2015). Although smaller in magnitude, other classes were also distinguished in terms of developmental correlates. For example, high verbal- and physical-group was associated with higher levels of aggression compared to the verbal-only group (Bradshaw et al., 2013). In sum, there seems to be both theoretical and empirical utility in distinguishing different types of bullying-victims. These subgroups displayed distinct prevalence rates and were associated with distinctive patterns of developmental problems, with groups in overlapping types of victimization showing the highest levels of negative correlates.

Applying Integrated L-RAT and Target Congruence Theories

We hypothesize that the integrated lifestyle-routine activities theory (Cohen & Felson, 1979; Cohen et al., 1981; Hindelang et al., 1978) and target congruence theory (Finkelhor & Asdigian, 1996) can help unpack and explain the subcategories of bullying-victims. Lifestyle-routine activities theory situates the risk of victimization with the lifestyle of victims, suitability of targets, availability of capable guardians, and proximity to motivated offenders. The theory suggests victimization is more likely among individuals who are unguarded and have characteristics that are attractive to potential offenders (Cohen & Felson, 1979; Cohen et al., 1981; Hindelang et al., 1978). Target congruence theory extends lifestyle-routine activities theory by pointing to specific aspects of victims that increase risk. Finkelhor and Asdigian (1996) summarized those characteristics in terms of vulnerability, gratifiability, and antagonism. In essence, there are personal attributes that make individuals more vulnerable targets by reducing their ability to deter perpetrators, making it more gratifiable to victimize them due to their intrinsic value, which ultimately instigates antagonism from the perpetrators by arousing negative emotions.

Central to the theory is the assumption that certain characteristics, such as physical weakness, negative emotionality, or social isolation are associated with reduced ability to resist peer perpetration and make individuals more vulnerable targets. For example, subsequent research has found positive association among psychological distress, depression, exposure to family violence, and anxiety with peer victimization risk (e.g., Frías & Finkelhor, 2017; Kahle & Peguero, 2017; Kulig et al., 2017; Zavala & Whitney, 2019). Also, individuals that possess the attributes that perpetrators want to have access to or find attractive may experience greater odds of victimization due to increased gratifiability. Target gratifiability has been applied to youth’s sex and produced mixed findings. Scholars have found positive association (e.g., Kulig et al., 2017; Zavala & Whitney, 2019), no relationship (e.g., Augustine et al., 2002), and negative link (e.g., Frías & Finkelhor, 2017; Kahle & Peguero, 2017) between being male and victimization. Lastly, target antagonism increases risk of victimization when potential victims are impulsive or aggressive and cause negative reactions from others (e.g., Augustine et al., 2002; Kulig et al., 2017). Finkelhor and Asdigian (1996) argue that these attributes are congruent with the needs and motives of perpetrators and thus increases the risk of victimization independent of potential victims’ lifestyles. To that end, the target congruence process should be conceptualized as a unique addition to lifestyle-routine activities theory and not as being part of the original theory. Therefore, we integrate target congruence variables with the expectation that those variables will contribute beyond the factors reflecting victims’ lifestyle and routine activities.

Does Target Congruence Theory Apply to Bullying-Victims?

There are two reasons to suggest that the integrated lifestyle-routine activities theory and target congruence approach can be applied to victims of bullying. First, target congruence theory was originally conceptualized to explain personal victimizations experienced by youth. While traditional street crimes such as robbery and stranger assault are well-explained by proximity to potential perpetrators (i.e., lifestyle), assaults derived from more personal motives are not completely explained with proximity to high crime areas or involvement in risky lifestyles. Rather, variables representing attributes that increase target attractiveness could be more relevant in explaining non-stranger attacks. For example, target congruence features were significantly related to adolescents' risk of sexual assault, parental assault, and assault due to bias (Finkelhor & Asdigian, 1996; Savolainen et al., 2020; Zavala & Whitney, 2019). A similar logic can be applied to the context of bullying, as incidents happen on school grounds where both victims and non-victims share close proximity to potential perpetrators. Risky lifestyles and proximity to offenders do not entirely explain victim selection, as some individuals get victimized without being involved in delinquency. It is more likely that the attributes that make targets more or less attractive to perpetrators can distinguish different types of bullying-victims.

Second, target congruence theory is an age-graded victimization theory. Power imbalance and antagonism are important components that are particularly relevant to bullying more so than other types of crime or deviant behavior. Especially in the case of school-based personal victimization, which include ridicule, teasing, and bullying based on personal attributes, target characteristics may be more relevant than situational or lifestyle factors (Henson et al., 2010). Indicators of target congruence (e.g., negative health status, sex, physical appearance, experience of violence in family, or weak strength) were positively associated with bullying victimization beyond the impact of peer delinquency, parental guardianship, or proximity to crime among adolescents (Augustine et al., 2002; Due et al., 2005; Frías & Finkelhor, 2017; Kahle & Peguero, 2017; Kulig et al., 2017; Stutzenberger, 2020). It is on these grounds that the propositions of target congruence are applicable to victims of bullying.

Research Context: Examining Bullying Victimization in South Korea

Bullying is a significant social problem that has recently received a significant attention in South Korea. Addressing violence in school has consistently been one of the priorities of previous and current government administrations. The School Violence Prevention and Countermeasure Act was introduced in 2004 and was subsequently expanded drastically to take on more proactive measures to prevent bullying and support victims. The Act requires that prevention councils and committees be established at the national and local levels. According to the national statistics on bullying, the largest portion comprised of victims of verbal bullying (35.6%), followed by group-based bullying (23.2%) and physical bullying (8.6%) (Korean Ministry of Education, 2019). Kim and colleagues (2016) reported that among their sample of 1640 middle school youths, 5% experienced physical bullying while 23% experienced verbal and relational bullying victimization (Kim et al., 2016). Still, little is known with regard to overlapping victims in Korea.

Current literature has documented various risk factors associated with bullying victimization among South Korean youths. Compared to those who are not exposed to bullying, victims tend to report higher levels of depressive symptoms, anxiety, body dissatisfaction, and lower levels of both academic achievement and self-esteem. Bully victims also did worse in school, had negative relationship with parents, and reported a lower socioeconomic status (Seo et al., 2017; Yang et al., 2006). These observations are in line with victimization literature that suggests bullying victimization is associated with behavioral, emotional, and social problems in non-Korean samples (e.g., Borowsky et al., 2013; Due et al., 2005; Kahle & Peguero, 2017). It appears that youth who are seen by their peers as vulnerable in terms of physical and psychological well-being are more likely to become targets of social exclusion, which can subsequently lead to worse physical and mental health.

Researchers have applied an integrated approach using personal attributes and situational theories to explain bullying victimization risks among Korean youths. The results have been generally supportive of the mediating role of risky lifestyle between individual traits and victimization risks. Certain individual traits (e.g., low self-control) were predictive of increased risk of victimization by increasing exposure to potential offenders and risky situations. Korean youths who had lower levels of self-control were more likely to be victimized, as well as to associate with delinquent peers which led to an increased risk of peer victimization (Cho & Wooldredge, 2018; Cho, 2018). Furthermore, youths with steeper growth rates of delinquent peer association were more likely to engage in deviant behavior and experience victimization (Cho, 2019). These studies contributed to the literature by demonstrating the applicability of an integrated theoretical approach in identifying the risk and/or protective factors associated with victims and non-victims of bullying.

In summary, bullying in South Korea takes on different subtypes and poses a significant threat to youths’ well-being. Yet less is known whether some subtypes are likely to co-occur with other types. Existing research on Korean youths has demonstrated that personal attributes and opportunity variables are relevant to risk of bullying victimization beyond the Western sample. The current study extends this knowledge area by investigating subtypes of victims and the factors associated with different groups of victims through the lens of target congruence and opportunity theories. Our study tests the utility of combining theoretical perspectives in distinguishing subtypes of bullying-victims. This integrative perspective distinguishes the present study from existing studies that applied either routine-activities theory to cyberbullying (e.g., Choi et al., 2019) or the target congruence perspective to a multicultural sample (e.g., Choi et al., 2020, 2021). Furthermore, this study examines the applicability of criminological models to a non-Western sample. The findings are expected to add evidence to cross-national generalizability of target congruence and opportunity theories.

Present Study

The current study employed a person-centered approach, which focuses on the multi-faceted dimensional aspect of bullying victimization to address methodological limitations in empirical research that use a variable-centered approach that oversimplifies the relations among individuals. In other words, this study determines the extent to which subtypes of bullying victimization co-occur among Korean youth. In addition, the study addresses theoretical limitations of lifestyle-routine activities theory by integrating target congruence theory for youth who are bullied without any involvement in risky lifestyles. The study takes up both of these theoretical underpinnings using an exclusively South Korean sample. Thus, this study extends prior victimization research in three ways by: (1) identifying subtypes of bullying victimization, (2) testing an integrated approach of target congruence theory and lifestyle-routine activities theory, and (3) classifying latent subgroups of Korean youth who may show different etiologies.

As suggested by the literature that the multi-faceted dimensions of bullying victimization were identified, we expect to find distinctive groups of South Korean youths with varying levels and subtypes of victimization (Hypothesis 1). We also argue that target congruence framework can be applied to distinguish subgroups of bullying-victims among South Korean youths. Compared to low or no victimization groups, youths with higher victimization rates should be associated with higher levels of target vulnerability, gratifiability, and antagonism. Based on Finkelhor and Asdigian (1996)’s conceptualization, target vulnerability should reflect the victim’s capability (or lack thereof) to resist victimization. We included depression, perceived stress, social withdrawal, and childhood maltreatment as indicators of youths’ vulnerability (Hypothesis 2-1). Target gratifiability refers to victims’ attributes that perpetrators may want to have access to or manipulate. Finkelhor and Asdigian (1996) included youth’s sex in the model and observed increased likelihood of victimization among males. Likewise, we expect to find greater proportion of males among higher victimization groups (Hypothesis 2-2). Target antagonism manifests characteristics that arouses jealousy, anger, negative reactions from perpetrators. Previous tests have included various attributes such as race, aggression, antagonistic behavior, or sexual orientation. We use aggressiveness in our model as our data did not contain youths’ sexual orientation and enough variation in youths’ race (Hypothesis 2–3). In addition to target congruence, we examine the relevance of bullying perpetration in distinguishing victim subtypes. Groups with higher levels of victimization are expected to show higher degrees of bullying perpetration relative to low or non victims (Hypothesis 3, 4).

Hypothesis 1

There would be distinct groups of Korean youths based on their profiles of past-year victimization regarding a subtype of bullying.

Hypothesis 2-1

Regarding target vulnerability, youth who have higher levels of depression, perceived stress, and social withdrawal, as well as those who experience greater child maltreatment, will have greater odds of being a member of a certain subtype of bullying victimization.

Hypothesis 2-2

Regarding target gratifiability, male youth will have greater odds of being a member of a subtype of bullying victimization.

Hypothesis 2-3

Regarding target antagonism, youth who demonstrate higher levels of aggressiveness will have greater odds of being a member of a subtype of bullying victimization.

Hypothesis 3

Regarding a risky lifestyle, youth who are involved in bullying perpetration will have greater odds of being a member of a subtype of bullying victimization.

Hypothesis 4

The significant relationship of target vulnerability, gratifiability, and antagonism will be reduced by bullying perpetration, conditional on a subtype of bullying victimization.

Methods

Sample

Data for the current study were obtained from the Korean Children and Youth Panel Survey (KCYPS), collected by the National Youth Policy Institute (NYPI) that was responsible for the data collection and all the fieldwork, including ethical considerations. The NYPI is a highly respected government agency that was established in 1989 and has functioned as a specialized research institution for youth policy studies since (see http://archive.nypi.re.kr/ for more information on the Institute and the dataset). The KCYPS is a nationally representative, longitudinal panel survey of Korean children and youth, utilizing a multi-stage stratified cluster sampling method where 151 elementary schools were selected in 17 national areas. These schools were randomly sampled proportionately to their size based on the average number of fourth grade students per class for the first wave (2010). Next, students and their parents were randomly sampled proportionately to the number of students enrolled in the selected schools. Standard ethical procedures for the responsible conduct of research were followed in both data collections during the first school visit to explain the goals of the survey, request cooperation, and provide an official letter for students and their parents, including active parental consent for child participation and student written assent related to human subjects’ protection. Parents also participated through a phone interview during the study; the study concluded with a thank you and a feedback-call to parents, also requesting continued participation for both the child and the parent.

The KCYPS contained information regarding various aspects of children and youths’ growth and development that were reported by the youths themselves, while socioeconomic status and family structure were self-reported by the parents. An in-person survey was given to students in the schools, whereas a telephone survey was administered to their parents or guardians. The sample for the current study was taken when the question of deviance began to be included from 2012 (the third wave). The participants in the current sample were in the sixth grade. The total sample size used for the analysis was 2217. For missing data, full informal maximum likelihood (FIML) was used, namely maximum likelihood estimation with robust standard errors (MLR), which is considered the state of art for handling missing data (Graham, 2012). Despite “MLR” techniques, cases with missing values on explanatory variables were excluded for some models. For this reason, a numerical iteration algorithm option was specified to include cases with missing values on all relevant explanatory variables in the models (Asparouhov & Muthén, 2010).

Measures

Dependent variable

Bullying victimization was assessed by asking respondents to indicate whether they had been bullied (1) physically (i.e., “collectively bullied,” “severely beaten,” and “taken away things away from being threatened”) and (2) verbally (i.e., “severely tested or bantered,” “threatened,” and “taunted or abused”) in the past year. These questions were constructed as a part of School Violence Victimization Questionnaire and have been found to have internal and construct validity (Kim et al., 2015; Lee et al., 2014). While scholars have used these items to measure bullying victimization (e.g., Choi et al., 2019; Hong et al., 2017; Lee, 2017), the cross-sectional nature of our study precluded us from accounting for repeated/continued occurrence of victimization. Those dichotomous scale items were used for the sake of parsimony in the latent class analysis.

Independent variables

Depression . The first independent variable capturing target vulnerability was depression. It was measured by using the following seven items: (1) “I am worn out and disheartened,” (2) “Sometimes I feel extremely sad and gloomy with no apparent reason,” (3) “I worry about everything,” (4) “Sometimes I cry,” (5) “I believe that everything will be due to me even when I get involved in trouble,” (6) “Sometimes I feel extremely lonely with no apparent reason,” and (7) “I feel low energy or slow down” (Radloff, 1977; Zavala, 2020).

Perceived stress . Perceived stress was captured by asking respondents the following two questions: “I get stressed by dissatisfaction over my future” and “I get stressed by hard work” (Choi et al., 2020).

Social withdrawal was captured using the following five questions: (1) “I feel really awkward when I’m in front of many people,” (2) “I feel embarrassed on many occasions,” (3) “I have difficulty expressing myself in front of others,” (4) “I feel shy,” and (5) “I don’t like to assert myself in front of people” (Hong et al., 2017). Response options for these three measures were provided using a four-point Likert scale, ranging from 1 = very untrue to 4 = very true. These measures demonstrated satisfactory internal reliability (α = .89 for depression; α = .82 for perceived stress; α = .89 for social withdrawal). We conducted a confirmatory factor analysis (CFA) with these items to derive these three measures of target vulnerability, respectively. Higher scores indicated higher levels of depression, perceived stress, and social withdrawal.

Child maltreatment was captured using: (1) child neglect (i.e., “Parents/someone in my family thinks that I am more important than their job,” “Parents/someone in my family are interested in my life,” “Parents/someone provide for my clean clothes, and “I knew that there was parents/someone in my family to take care of me and protect me”) and (2) child abuse (i.e., “The punishments I received seemed cruel,” “I was punished with a belt, a board, a cord, or some other hard object,” “Parents/someone in my family hit me so hard that it left me with bruises or marks,” and “Parents/someone in my family said hurtful or insulting things to me”). The items of child neglect were reverse coded, and the eight items of child neglect and abuse were coded as 1 = very untrue to 4 = very true. CFA was conducted to create a single latent of child maltreatment with higher scores reflecting higher likelihood of being abused and neglected by parents/someone.

Sex . Following the original conceptualization of target gratifiability used by Finkelhor and Asdigian (1996), we included youths’ sex in the model. Zavala and Whitney (2019) utilized the same operationalization and found significant impacts of sex on risks of victimization. In our sample, 52.7% were male and 47.3% were female.

Aggressiveness capturing target antagonism was conceptualized using the following six items developed by Jo and Im (2003) and modified by the National Youth Policy Institute: (1) “I found fault with little things,” “(2) I interrupted other people,” (3) “If someone didn’t let me do what I want, I argued with him/her,” (4) “I often got into arguments over nothing,” (5) “I sometimes got angry all day long,” and (6) “I sometimes cried for no reason” (Um et al., 2019). These items demonstrated satisfactory internal reliability (α = .84). Response options were a four-point Likert scale, ranging from 1 = very untrue to 4 = very true. We conducted CFA with the items to derive one measure of target antagonism with higher scores indicating higher levels of aggressiveness.

Bullying perpetration was used to measure a risky lifestyle, capturing the opportunity variable. This variable was captured by asking respondents to indicate whether they engaged in the following six bullying behaviors: “severely teasing or bantering other friends,” “collectively bullying,” “severely beating others,” “threatening others” “taking things away with threats,” and “stealing others’ items” in the past year. Those dichotomous scale items were summed to signify whether respondents had engaged in any of the bullying behaviors. Although the summed variable was a scale in the psychometric sense of multiple items accessing a single concept, it was recorded as a dichotomous variable (0 for no experience and 1 for the composite item). This gave the computed variable the best chance of accuracy to account for the item-level missing data. All items were added to Table 1 (including one with 0 = none and 1 = each deviance).

Table 1.

Descriptive Statistics (N = 2217).

Variable	Range	Mean	SD
Bullying victimization
Being severely teased or bantered during the last 1 year	1 = yes0 = no	0.05	0.23
Being collectively bullied during the last 1 year		0.03	0.18
Being severely beaten during the last 1 year		0.01	0.11
Being threatened during the last 1 year		0.01	0.12
Being taken things away from being threatened during the last 1 year		0.03	0.16
Being taunted/abused during the last 1 year		0.08	0.27
Target vulnerability
Depression (⍺ = 0.89)
I am worn out and disheartened	1 = very untrue2 = untrue3 = true4 = very true	1.81	0.80
Sometimes I feel extremely sad and gloomy with no apparent reason		1.72	0.79
I have many worries (I worry about everything)		1.90	0.89
Sometimes I cry		1.84	0.85
I believe that everything will be due to me, even when I get involved in trouble		1.86	0.84
Sometimes I feel extremely lonely with no apparent reason		1.64	0.78
I feel low energy or slow down		1.58	0.70
Perceived stress (⍺ = 0.82)
I get stressed by dissatisfaction over my future		1.57	0.74
I get stressed by hard work		1.60	0.76
Social withdrawal (⍺ = 0.89)
I feel really awkward when I’m in front of many people		1.86	0.89
I feel embarrassed on many occasions		2.15	0.94
I have difficulty expressing myself in front of others		1.97	0.89
I feel shy		2.02	0.91
I don’t like to assert myself in front of people		2.21	0.92
Child maltreatment (⍺ = 0.78)
Parents/someone in my family thinks that I am more important than their job	1 = very true2 = true3 = untrue4 = very untrue	1.78	0.89
Parents/someone in my family are interested in my life		1.68	0.73
Parents/someone provide for my clean clothes		1.56	0.67
I knew that there was parents/someone in my family to take care of me and protect me		1.50	0.66
The punishments I received seemed cruel	1 = very untrue2 = untrue3 = true4 = very true	2.07	0.83
I was punished with a belt, a board, a cord, or some other hard object		1.59	0.72
Parents/someone in my family hit me so hard that it left me with bruises or marks		1.54	0.77
Parents/someone in my family said hurtful or insulting things to me		1.52	0.78
Target gratifiability
Sex (male)	1 = males0 = females	0.53	0.50
Target antagonism
Aggressiveness (⍺ = 0.84)
I found fault with little things	1 = very untrue2 = untrue3 = true4 = very true	2.15	0.80
I interrupted other people		2.06	0.79
If someone didn’t let me do what I want, I argued with him/her		2.01	0.81
I often got into arguments over nothing		2.10	0.82
I sometimes got angry all day long		2.09	0.89
I sometimes cried for no reason		1.78	0.83
Opportunity
Bullying perpetration
Severely teasing or bantering other friend during the last 1 year		0.03	0.18
Collectively bullying during the last 1 year		0.04	0.19
Severely beating other people during the last 1 year		0.01	0.11
Threatening other friend during the last 1 year		0.01	0.10
Taking things away with threats during the last 1 year		0.01	0.08
Stealing during the last 1 year		0.01	0.07
Family SES
Father’s level of educational attainment	1 = low or middle school2 = high school3 = 2-year college4 = undergraduate5 = graduate	3.11	1.07
Mother’s level of educational attainment		2.89	1.00
Father’s job title(s)	1 = laborer, production worker2 = sales, service worker3 = clerical staff4 = owner of a business, professional	2.31	1.20
Mother’s job title(s)		2.64	1.07
Monthly household income	1 = 1000 or less2 = 1001 to 3,0003 = 3001 to 5,0004 = 5001 to 7,0005 = 7001 to 9,0006 = 9001 or more	3.10	1.09

Note. Sex is reported as a percentage.

Control variable

Socioeconomic status (SES) of family was captured by using five items¹: parents’ level of educational attainment, job title(s) of parents, and monthly household income. Parents’ level of educational attainment had an average of 3.11 (father; SD = 1.07) and 2.89 (mother; SD = 1.00) in the range of 1 (low or middle school) to 4 (graduate school). The job titles parents had were an average of 2.31 (father; SD = 1.20) and 2.64 (mother; 1.07) in the range of 1 (laborer and production worker) to 4 (owner of a business and professional). Annual household income had an average score of 3.10 (SD = 1.09), ranging from 1 ($1000 or less) to 6 ($9001 or more). All the items led to emerging the three types into a composite scale with higher scores indicating higher levels of family SES. Descriptive statistics and survey items of all variables for the analysis are shown in Table 1. Age and race were not included as a control variable. All participants were in the same seventh grade, and approximately two percent of students enrolled in elementary school identified as having multicultural backgrounds (Korean Ministry of Education, 2020), making the population racially homogeneous.

Analytic Strategy

Using Mplus version 7.4, the current study employed latent class approach to identify the heterogeneity in bullying victimization among Korean youth, examining how individual trait characteristics and risky lifestyles were differentially associated with patterns of bullying victimization (Muthén & Muthén, 2015). The observed sample of Korean youth were classified into subgroups that exhibited similar patterns of responses in individuals (i.e., class homogeneity) and dissimilarities across subgroups (i.e., class separation) rather than shared covariance among measures.

The analysis of this study utilized a manual 3-step approach. The first step was determining the optimal number of classes, each exhibiting homogeneity in bullying victimization (the class enumeration process). This study ran a series of models to determine the final optimal class model by ascending the number of specified classes from 1 to 6. The final optimal solution was evaluated based on a comparison among multiple statistical indices (i.e., model fit indices) such as absolute model fit, including the model maximum log likelihood (LL) value and an overall likelihood ratio model chi-square goodness-of-fit (X²_LRT) and relative model fit, including AIC, BIC, SABIC, CAIC, and AWE (see the bottom of Table 2). Additionally, for likelihood ratio tests (LRT) statistics, both the Lo-Mendell-Rubin likelihood ratio test (LMR-LRT) and the bootstrapped LRT (BLRT) were used to compare two adjacent/nested models in assessing whether a k-1-class model (H₀) provided a better fit than a kclass model (H₁). For LMR-LRT and BLRT, the significant p-value indicates that the K-class (H₁) model provides better fit than the k-1 class (H₀) model (Tein et al., 2013). Last, Entropy² was used, ranging from 0 to 1, with higher values representing better classification. The second step consisted of a class modification process with measurement error variances (i.e., misclassification rates/uncertainty rates). Latent class analysis allows for uncertainty rates in class-specific item probabilities that might lead to biased estimates and standard errors for the relationship of class memberships with other background variables. In other words, adding other background variables to subsequent models might affect the classification and interpretation of classes because additional information obtained by the added variables might influence the classification process (Lubke & Muthén, 2007). For this reason, this study estimated a nominal most likely class model, creating a nominal indicator of the categorical classes. The nominal variable (the class-specific multinomial intercept) was prefixed at the misclassification error variances (i.e., logit values, compiled from Mplus outputs of the final optimal solution model, entitled “Logits for the Classification Probabilities for the Most Likely Latent Class Membership (Column) by Latent Class (Row)”). Thus, the classification and its interpretation were not changed by the subsequent addition of variables. In the last step, multinomial logistic regression analyses were conducted to examine the relationship with youth’s trait characteristics and risky lifestyles on latent class membership. Two main models were tested, namely (1) youth’s trait characteristics (sex, aggressiveness, depression, perceived stress, social anxiety, and child maltreatment) and (2) youth’s trait characteristics and risky lifestyles (full model). Zero-order relationships between each predictor and other variables are presented in Table 2.

Table 2.

Correlations among the Study Variables.

Variable	1	2	3	4	5	6	7	8	9	10	11	12	13	14
1	−	.322^**	.220^**	.203^**	.139^**	.355^**	.189^**	.129^**	.117^**	.087^**	.023	.124^**	.150^**	.013
2		−	.154^**	.162^**	.143^**	.232^**	.110^**	.077^**	.054^*	.047^*	−.098^**	.050^*	.205^**	.027
3			−	.412^**	.207^**	.202^**	.050^*	.031	.004	.048^*	.030	.053^*	.104^**	−.050^*
4				−	.215^**	.296^**	.080^**	.046^*	.010	.038	.020	.072^**	.108^**	−.039
5					−	.217^**	.084^**	.060^**	.032	.033	−.029	.068^**	.151^**	−.019
6						−	.152^**	.114^**	.066^**	.110^**	.033	.113^**	.219^**	−.012
7							−	.783^**	.611^**	.383^**	−.143^**	.474^**	.128^**	−.025
8								−	.448^**	.423^**	−.078^**	.410^**	.112^**	−.027
9									−	.218^**	−.076^**	.321^**	.083^**	−.008
10										−	.085^**	.328^**	.077^**	−.045^*
11											−	.039	−.020	−.036
12												−	.121^**	−.001
13													−	.047^*
14														−

Notes. 1 = being severely teased or bantered; 2 = being collectively bullied; 3 = being severely beaten; 4 = being threatened; 5 = Having been taken things away from being threatened; 6 = being taunted or abused; 7 = depression: 8 = perceived stress; 9 = social anxiety; 10 = child maltreatment; 11 = sex (male); 12 = aggressiveness; 13 = bullying perpetration; 14 = family SES.

*p < .05; ** p < .01; *** p < .001.

Results

Patterns of Bullying Victimization

Latent class analysis was conducted to determine the optimal latent class model of bullying victimization with one through six classes specified. The absolute and relative model fit indices, as well as the classification diagnostics, were used to empirically determine the final optimal solution model (see Table 3). Among the four models, the 4-class model appeared to best fit the data with the lowest LL and AIC values, while the 2-class model had the best model fit to the data with the lowest BIC, SABIC, and AWE. However, both adjust LMR-LRT and BLRT of the 4-class model provided a non-significant p-value (p = .24 and .14, respectively), indicating that we failed to reject the k-1 (H₀) class, the 3-class model. This means that the 3-class model that provided the best fit to the observed patterns of response was chosen as the optimal solution. Since there is no clear-cut decision rule, multiple model fit statistics should be paired with a theoretical understanding of the constructs. Table 4 presents posterior probabilities of assignment to the latent classes in the chosen model, meeting criteria for fit of the posterior probabilities for most-likely latent class membership. The final optimal solution yielded average posterior probabilities, ranging between 0.919 and 0.959 (above the 0.70 threshold, suggested by Nagin, 2005), indicating the chosen mode was successful in assigning Korean youth to the classes with good classification quality. With these criteria, we concluded that the 3-class solution model was most acceptable.

Table 3.

Model Fit Indexes for Class Enumeration of South Korean Youth Bullying Victimization Items.

K	LL	n-par	LR X₂	df	p-value	AIC	BIC	SABIC	AWE	H0: K-1 classes; H1: K classes
K	LL	n-par	LR X₂	df	p-value	AIC	BIC	SABIC	AWE	LRTS	Adj LMR p-value	Boot strapped p-value	Entropy
1-class	−1991.97	6	303.23	48	<.001	3995.94	4030.17	4011.10	4088.33
2-class	−1736.62	13	80.21	50	<.001	3499.23	3573.39	3532.08	3699.41	501.41	<.001	<.001	0.90
3-class	−1721.78	20	50.55	43	.2	3483.57	3597.65	3534.10	3791.53	29.12	<.01	<.001	0.89
4-class	−1714.00	27	34.98	36	.52	3482.00	3636.01	3550.23	3897.74	14.99	.24	.14	0.92
5-class	−1711.95	34	30.87	29	.37	3491.90	3685.83	3577.81	4049.76	3.208	.24	1.00	0.89
6-class	−1710.51	41	28.01	22	.18	3503.03	3736.89	3606.63	4175.75	1.800	.35	1.00	0.63

Note. N = 2217. Bold numbers indicate fit indices of the selected models. LL = Model maximum log likelihood value (the value shown in bold indicates a model with the smallest LL value that perfectly fits the data); npar = number of free parameters estimated in the model; LR X₂ = likelihood ratio model chi-square goodness-of-fit (the value shown in bold indicates the model with the smallest LR X₂ fits observed data); p-value (the non-significant p-value indicates that the null hypothesis may not be rejected, showing a better model fit to observed data); AIC = the Akaike Information Criterion (the value shown in bold indicates the model with the smallest value); BIC = the Bayesian Information Criterion (the value shown in bold indicates the model with the smallest value); SABIC = the sample-size adjusted BIC (the value shown in bold indicates the model with the smallest value); AWE = Approximate Weight of Evidence Criterion (the value shown in bold indicates the model with the smallest value); LRTS = likelihood ratio test statistics comparing a current model (k class) to a model with one less latent class (k-1 class); Adj LMR p-value = the adjusted Lo-Mendell-Rubin likelihood ratio test p-value (the value shown in bold represents the non-significant p-value, indicating the current model with the smaller number of classes is not rejected; the significant p-value indicates the k-class model provides significantly better fit to the data than the k-1 class model); Bootstrapped p-value = parametric bootstrapped p-value for the LRTS; Entropy = represented excellent classification.

Table 4.

Posterior Probabilities of Assignment to Latent Trajectory Classes.

	Low/Non Victims (91.2%)	Physical & Verbal Bullying Victims (0.5%)	Verbal Bullying Victims (8.3%)
Low/Non victims	0.96	0	0.04
Physical & verbal bullying victims	0	0.92	0.08
Verbal bullying victims	0.06	0.02	0.93

Note. Average Latent Class Probabilities for Most Likely Latent Class Membership (Row) by Latent Class (Column).

The item-response probabilities of the optimal 3-class model are displayed in Table 5. The probabilities corresponding to the “yes” response were used to interpret the meaning of the latent classes and assign labels to the latent classes. Youths in class 1 had the lowest probabilities associated with the “yes” response for all six items, suggesting a label of Low/Non-Victims. Youths in class 2 had the highest probabilities corresponding to the “yes” response for the five items (physical and verbal bullying victimization) and were thus labeled as Physical & Verbal Bullying Victims. Youths in class 3 had medium-high probabilities associated with the “yes” response for one item (i.e., being taunted or abused), and were subsequently labeled as Verbal Bullying Victims. Figure 1 shows the overall pattern of the three classes.

Table 5.

Item Response Probabilities and Membership of Three-Latent-Class Model (N = 2217).

Assigned label	Latent class of bullying victimization
	1	2	3
	Low/Non Victims	Physical & Verbal Bullying Victims	Verbal Bullying Victims
Probability of membership	91.2%	0.5%	8.3%
Item-response probabilities	Probabilities of responding yes*
Being severely teased or bantered	0.02	0.80**	0.42
Being collectively bullied	0.01	0.48	0.25
Being severely beaten	0.00	0.82**	0.09
Being threatened	0.00	0.96**	0.10
Having been taken things away from being threatened	0.01	0.65**	0.15
Being taunted or abused	0.03	0.96**	0.59**

Notes. *Recoded from original response categories. **Item-response probabilities >.5 in bold to facilitate interpretation. The probability of a “No” response can be calculated by subtracting the item-response probabilities shown above from 1. Memberships across statuses may not sum up to 1 (100%) due to rounding.

Figure 1.

Probability of Bullying Victimization based on Class Membership (N = 2217).

Class Membership by Background Variables

Results of the multinomial logistic regression analyses are reported in Table 6. Statistically significant logit coefficients of class 2 (Physical & Verbal Bullying Victims) and class 3 (Verbal Bullying Victims) were detected in comparison with Low/Non-Victims as the reference group (in Model 1 and Model 2, respectively). In Model 3, logit coefficients of class 2 (Physical & Verbal Bullying Victims) were estimated, compared to class 3 (Verbal Bullying Victims).

Table 6.

Latent Class Regression Model Estimating Target Congruence and Lifestyle Effects on Class Membership in Bullying Victimization.

	Model I						Model II
	Estimation of target congruence effects						Estimation of target congruence + lifestyle effects
	Model 1		Model 2		Model 3		Model 1		Model 2		Model 3
	Physical & verbal bullying victims versus Low/non victims		Verbal bullying victims Low/non victims		Physical & verbal bullying victims versus Verbal bullying victims		Physical & verbal bullying victims versus Low/non victims		Verbal bullying victims versus Low/non victims		Physical & verbal bullying victims versus Verbal bullying victims
	Est	SE	Est	SE	Est	SE	Est	SE	Est	SE	Est	SE
Control variable
Family SES	-.91***	.19	.28	.23	−1.19***	.31	−.92***	.25	.23	.23	−1.15***	.33
Target vulnerability
Depression	−.87	.55	1.29***	.37	−2.16**	.71	−1.68	.92	1.21***	.37	−2.88**	1.03
Perceived stress	.97	.60	−.40	.30	1.37	.75	1.47	.91	−.37	.31	1.85	1.03
Social anxiety	.29	.58	−.28	.26	.57	.69	.41	.75	−.18	.24	.59	.80
Child maltreatment	−.21	.43	.31	.17	−.51	.47	−.26	.38	.25	.18	−.52	.40
Target gratifiability
Sex (male)	.37	1.33	.18	.36	.18	1.41	.70	1.20	.22	.37	.49	1.25
Target antagonism
Aggressiveness	.46*	.23	.44*	.20	.02	.32	.77**	.28	.40	.23	.37	.35
Risky lifestyle
Bullying perpetration							3.70*	1.68	1.75***	.41	1.95	1.73
Model fit information
Akaike (AIC)	487.94						468.32
Bayesian (BIC)	569.35						559.91
Sample-size adjusted BIC	518.53						502.74

Note. Boldface entries are unstandardized coefficients and reflect statistically significant coefficients (* p < .05; ** p < .01; *** p < .001).

Parameterization using Reference Class 1 (low/non victims).

This study estimated class uncertainty rates (i.e., misclassification error variances) of the optimal 3-lass model. Before specifying the background variables, this study compiled information about classification quality (i.e., the logit values of classification uncertainty rates) by obtaining the individual posterior probability of each of the three classes to use in a subsequent model. Then, we created a most likely class variable, which is a nominal indicator of the three classes. The nominal indicator were prefixed at the logit values of each class, obtained from the mplus output of the optimal 3-class model. By doing so, the latent class classification in the unconditional model was held constant so that the class membership was not influenced by the subsequently added background variables.

Model I of Table 6 tested the association among target vulnerability, gratifiability, and antagonism and the likelihood of bullying victimization. In Model 1, aggressiveness was significantly and positively related to class 2 membership, compared to the reference group (β = 0.46, p < .05), indicating that youth who were aggressive were more likely to be victims of physical and verbal bullying compared to low/non-victims. Family SES was significantly and negatively related to class 2 membership, compared to the reference group (β = −0.91, p < .001). Youth who reported lower levels of family SES were more likely to be victims of physical and verbal bullying compared to low/non-victims. In Model 2, depression and aggressiveness were significantly and positively associated with class 3 membership, compared to the reference group (β = 1.29, p < 0.001; β = 0.44, p < .05, respectively). Youth who were depressed and aggressive were more likely to become victims of verbal bullying, compared to low/non-victims. In Model 3, depression was significantly and negatively related to class 2 membership, compared to the reference group (β = −2.16, p < .01), indicating that the log odds of victims of verbal bullying relative to victims of physical and verbal bullying were greater for youth who felt depressed than those who did not. Family SES was significantly and negatively related to class 2 membership, compared to class 3 (β = −1.19, p < .001). The log odds of victims of physical and verbal bullying relative to victims of verbal bullying were greater for youth who reported lower levels of family SES than those who did not.

The risky lifestyle factor, bullying perpetration was added in Model II of Table 6 to examine how it was associated with class membership, as well as whether the significant relationship of target vulnerability and antagonism and the class membership remained stable or were reduced in the full model. In Model 1 of Model II, bullying perpetration was significantly and positively related to class 2 membership, compared to the reference group (β = 3.70, p < .05). Youth who bullied others were more likely to be physically and verbally bullied. Aggressiveness was still significantly and positively related to victims of physical and verbal bullying (β = 0.77, p < .01). The log odds of victims of physical and verbal bullying relative to low/non-victims were greater for youth with higher aggressiveness. Family SES was still significant, indicating that youth with lower levels of family SES were more likely to be victims of physical and verbal bullying (β = −0.92, p < .001). In Model 2, bullying perpetration was significantly and positively related to class 2 membership, compared to the reference group (β = 1.75, p < .001). Youth who bullied others were more likely to be victims of verbal bullying. Depression remained significant (β = 1.21, p < .001), but was slightly reduced; while aggressiveness was nonsignificant. However, there was no significant difference in bullying perpetration between Classes 2 and 3 in Model 3, indicating that bulling perpetration was significant in both groups. Depression and family SES were still significant (β = −2.88, p < .01; β = −1.15, p < .001). The odds of being victims of physical and verbal bullying relative to those of verbal bullying were greater for youth who reported lower levels of family SES. However, the odds of being victims of verbal bullying relative to those of physical and verbal bullying were greater for youth who felt depressed.

Discussion

Many studies have found empirical support to suggest the significant association among individual traits, risky lifestyles, and youth victimization (Cho, 2017b, 2018; Cho & Lee, 2021; Moon & Alarid, 2015). It is uncertain, however, if the same framework is applicable to explaining direct and overlapping forms of bullying victimization. The purpose of this study was to assess the generality of the integrated approach of target congruence and lifestyles and routine activities theories and, substantively, whether this approach could be used to explain various forms of bullying victimization among a national sample of South Korean youth. Prior research has employed a relational, variable-oriented approach in which links among variables for all individuals are emphasized, potentially leading to causal over-simplification. Thus, this study sought to expand the extant empirical literature on bullying victimization by identifying multi-faceted dimensions of bullying and classifying Korean youth into distinct subgroups, each exhibiting unique patterns that may be associated with differential etiologies. The results produced several interesting findings.

First, we uncovered different patterns of victimization, including physical and verbal, verbal only, and low victimization. We tested varying patterns and numbers of classes on our sample of 2217 adolescents and, based on the p-value of LR X², LMR-LRT, and BLRT, we determined that the 3-class model provided the best fit to the data. Thus, we concluded that the data supported the first hypothesis by confirming multiple classes with heterogeneous types and probabilities of bullying victimization among teenagers. The majority of the sample (91.2%) belonged to the group with very low probabilities of reporting any type of victimization measures included in the data. A small group of youth (0.5%) reported the highest probabilities across all six measures of bullying victimization. Their probabilities of victimization exceeded 0.5 on all items except for collective bullying, which still reached almost half at 0.479. These items included both physical forms (e.g., being severely beaten) and verbal forms (e.g., being threatened). The last group accounted for 8.3% of the sample and was represented by reporting a medium-high probability of being verbally taunted and abused. These results were in line with those of previous studies in which distinct patterns of bullying victimization were found, and thus supported our first hypothesis. Existing studies that utilized latent class analysis commonly identified a large group of no/low level victims (approximately 50–85%), a small group of youths who had high probabilities of reporting all forms of bullying, and one or two remaining groups with overlapping forms of victimization (Berkowitz et al., 2015; Bradshaw et al., 2013, 2015). Although there is less consensus on the total number of classes across studies, our finding adds importance in acknowledging subgroups of bullying-victims with distinct patterns.

Our subgroups diverged from previous studies, however, on specific compositions within subgroups. Our low/no victimization group was larger compared to those identified in previous research, and a group of verbal-only victims was uniquely discovered in our analysis. The larger prevalence of low victim class can be explained by the nature of our sample. Previous studies relied on middle school and high school students while our sample consisted of relatively younger youths enrolled in elementary schools. Studies indicate that reports of bullying are lower in early adolescence, increase gradually to peak form in later adolescence, followed by a decrease after puberty (Pellegrini & Long, 2002; Pepler et al., 2006). Additionally, in contrast to earlier studies, we detected a class of verbal-only victims. This finding is distinguished from previous research in which verbal victims tend to report overlap with relational or physical forms. This discrepancy could be attributed to the pool of victimization questions used in the latent class analysis.

Next, we tested whether target congruence theory could distinguish different subgroups of bullying-victims. As indicated in the estimates of latent class regression Model 1, target vulnerability and antagonism variables distinguished non-victims and victims as well as physical and verbal victims against verbal-only victims, supporting Hypothesis 2-1, 3. Compared to low/non-victims, physical and verbal or verbal-only victims were more aggressive to others. Verbal-only victims reported higher levels of depression relative to the other two groups. These findings are consistent with previous studies in which bullying-victims reported higher internalizing problems and aggression compared to non-victims (Bradshaw et al., 2013, 2015). Juvonen and colleagues (2003) observed a similar finding in which youths who had emotional/behavioral problems were more likely to elicit negative interactions from peers. However, target gratifiability, measured along youths’ sex, did not significantly predict class membership of bullying-victims (Hypothesis 2-2). Regardless of sex, higher aggression seemed to make youths more vulnerable to overall bullying victimization, while being depressed increased the odds of being verbally bullied, partially supporting the notion of target congruence.

Finally, our full model (Model II) tested whether the target congruence variables would be relevant when deviant lifestyle was controlled for. Bullying perpetration was positively associated with the likelihood of being in one of the victim groups, consistent with Hypothesis 3. It did not, however, differentiate between physical and verbal and verbal-only groups. Youths who bullied others were more likely to be bullied in certain forms. Inclusion of bullying perpetration resulted in reduced effects of two target congruence variables, partially supporting Hypothesis 4. In Model 2 of Model II, the effect of depression was reduced (still positively and significantly related) and the effect of aggressiveness became nonsignificant after controlling for bullying perpetration. Youths who reported greater levels of depression and bullying behavior were more likely to be verbally abused. The association between bullying victimization and target congruence variables slightly increased in other models of Model II. Depression was still relevant in distinguishing physical and verbal victims and verbal-only victims in Model 3. Aggressiveness became increasingly important in distinguishing physical and verbal victims from non-victims (Model 1).

The relationship between family SES and class membership largely remained the same in Model II. Frías and Finkelhor (2017) also found that Mexican youths with lower SES had a higher likelihood of experiencing bullying victimization. Other studies, however, did not find family SES to be impactful in predicting violent or bullying victimization (Augustine et al., 2002; Kahle & Peguero, 2017). It is possible that the effect of SES is intertwined with that of race or minority status. Both the current analysis and Frías and Finkelhor’s (2017) study included a racially homogeneous sample while other studies did not. These differences suggest a need for additional research that addresses race/minority status and SES of youth.

In sum, our findings suggest that there are subtypes of bulling victimization with and without overlapping dimensions. The results generally supported applying the integrated perspective of target congruence and lifestyle-routine activities theory in explaining subtypes of bullying victimization among youths. Vulnerability and antagonism attributes were particularly relevant in distinguishing non-victims, victims of verbal-only bullying, and victims of physical and verbal bullying. Previous research that included both lifestyle-routine activities factors and target congruence variables had reported similar findings. Indicators of lifestyle-routine activities theory, measured by risky behavior, exposure to delinquency, and bullying perpetration, were significantly associated with violent or bullying victimization. Still, when these indicators were controlled for, target congruence variables (such as psychological distress, depression, social withdrawal) significantly predicted victimization (Augustine et al., 2002; Due et al., 2005; Finkelhor & Asdigian, 1996; Frías & Finkelhor, 2017; Kahle & Peguero, 2017; Zavala & Whitney, 2019).

Implications

The current study makes a theoretical contribution by demonstrating the utility of combining target congruence variables with lifestyle-routine activity concepts in explaining heterogeneity in victims of bullying. Target congruence variables were helpful in distinguishing subtypes of bullying-victims beyond the predictive power representing deviant lifestyles. We predicted that integrating target congruence and lifestyle-routine activities variables would be useful in our study exploring adolescent victims ofbullying on two grounds: (1) target congruence is an age-graded victimization theory particularly relevant for youths (Butler et al., 2019), and (2) target congruence holds a stronger predictive power for personal victimization compared to impersonal victimization. Bullying victimization satisfied both criteria and the results of our study supported the proposition. We confirmed Finkelhor and Asdigian’s (1996) claim that the concept of target congruence has an additive and independent value to lifestyle-routine activities theory. In line with their argument, the process that makes an individual a vulnerable target is more complex than one’s lifestyle and should account for personal attributes such as psychological distress, individual traits, and social competence (Finkelhor & Asdigian, 1996).

More importantly, the current study examined the applicability of target congruence and opportunity variables to subtypes of bullying-victims in South Korea. We found groups of victims who concurrently experienced physical and verbal bullying and those who primarily experienced verbal bullying. While confirming the existence of different subtypes of bullying among South Korean youths evidenced in previous investigations (e.g., Kim et al., 2016; Korean Ministry of Education, 2019), our findings added meaningful contribution by discovering the victims who experienced overlapping types. We were also able to test whether the integrated perspective is applicable to South Korean youths. Our findings indicated a partial support for the target congruence and a full support for the risky lifestyle perspective: depression, aggressiveness, and bullying behavior were highly relevant in distinguishing victims groups while other target congruence variables were not. Target congruence perspective has been tested with varying sets of variables and produced divergent findings. For example, psychological distress and single-parent family household were emphasized in Finkelhor and Asdigian (1996), experience of violence in family was impactful in Frías and Finkelhor (2017), and in Kahle and Peguero (2017), perceived support was emphasized, and family affluence did not show a significant association. Zavala and Whitney (2019) found relevance of anxiety, sex, and anger, but no significant relationship with depression, school performance, age, immigration or sexual orientation. The current study was the first attempt to apply target congruence and risky lifestyles perspective to distinguish victim groups among South Korean youths, and it is not clear whether cultural differences produced divergent results. Nevertheless, our findings show that it is at least partially applicable to non-Western sample and call for continued research.

Our findings can also be used to inform prevention and intervention strategies used to address bullying. First, teachers and relevant stakeholders should be aware of overlapping victims. Our analyses revealed latent groups of overlapping victims (i.e., physical and verbal) as well as non-overlapping victims (i.e., verbal-only). These groups were distinct in terms of depressive symptoms and SES. An effective intervention should assess the possibility of victimization subtypes. Second, bullying behavior was a robust and constant predictor of bullying-victimization for both the physical and verbal victims and verbal-only victims. Interventions should be directed toward youths who engage in delinquent behaviors as they are at higher risk of victimization. Students should be informed that their bullying acts can make them vulnerable for bullying victimization. Combined with previous research indicating reciprocal effect between bullying perpetration and victimization (e.g., Jose et al., 2012; Park & Cho, 2021; Turanovic & Pratt, 2014), our findings reinforce the need to address the dynamic relationship between being a victim and a bully.

Lastly, compared to non-victims, victims of physical/verbal or verbal bullying had lower SES, were depressed, and more aggressive. Individuals who are depressed or aggressive may be less competent in successfully socializing with peers and may be more likely to provoke or be subjected to social conflict, which could lead to bullying victimization. Research recommends including interventions that target small groups of youths who are more vulnerable to bullying-victimization (Gaffney et al., 2019). Target congruence features, especially low SES, depression, and aggression can guide the process of identifying subgroups that may need special attention in intervention. Intervention and treatment strategies should address victims’ underlying psychological correlates, particularly depression, in addition to observable bullying behaviors, as depressed children are found to benefit the least from anti-bullying interventions (Nocentini et al., 2019).

Limitations

Although this study contributed several key advancements to the aforementioned theoretical and methodological frameworks, there are three notable methodological shortcomings that are worth acknowledging. First, the primary measure of bullying victimization was assessed using 6 items except those involved in social/relational and cyber-bullying because the KCYPS did not contain indirect forms of bullying. Future studies would benefit from considering additional bullying indicators such as spreading rumors and cyberbullying. Although there is certainly no uniform definition of bullying, common themes in the literature suggest bullying is defined as aggression that intentionally occurs over time, in a repeated fashion between more powerful youths and less powerful youths (Olweus, 1993). As defined in the study, if students reported one occurrence during the previous year, they were identified as a bully-victim. While we were able to capture imbalance in power between victims and perpetration, this measure might lead to a lack of construct or predictive validity at face value. Future research would benefit from including other items that assess intensity, frequency, duration, or type of bullying behavior.

Second, we could only include a subset of target congruence variables and were limited from testing all concepts discussed by Finkelhor and Asdigian (1996) and subsequent tests. For example, our data did not contain certain information (e.g., physical size, race, or sexual orientation). While age is a theoretical relevant variable (i.e., target gratifiability), we could not control for it as our sample was identical in age. Still, we note that our selection of variables was made in accordance with theoretical definition of target congruence along with cultural relevance. Also, this study was only capable of assessing the effects of individual-level predictors in explaining class membership of bullying victimization in a cross-sectional format. Thus, we could not assume time-ordered relationships among study variables. Also, it remains possible that a period of time was captured where influences of target suitability and risky lifestyles were age-graded within a dynamic socialization process. Future research should incorporate a longitudinal design to accurately examine this dynamic process and establish causal relationships. Also, our aggressiveness scale was constructed with six items including “I sometimes cry for no reason.” Although the scale was empirically validated by the original developers, it is still possible that crying behavior does not conceptually align well with aggression. Our further analysis of the sample indicated that this item converged well with the other five items onto a single aggression factor, but a careful examination is still required to the researchers who use the same data.

Finally, this study examined the effects of individual traits and risky lifestyles. Individual-level opportunity factors on victimization can be reinforced or attenuated by macro-level opportunity factors. Environmental and contextual factors of schools (e.g., school type and size, classroom structure) and neighborhoods (high-crime area, socially disorganized area) might influence individual-level opportunities for interaction between students. For this reason, it would be better for future efforts to aggregate the data to higher levels (i.e., school or neighborhood) which would yield enough cases to conduct different levels of analysis (Bryk & Raudenbush, 1992).

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

ORCID iDs

Sujung Cho

Insun Park

Notes

Author Biographies

Sujung Cho is an assistant professor in the Department of Criminology and Criminal Justice at Southern Illinois University Carbondale. She is working on developmental research, which presupposes developmental determinism where childhood delinquent behavior leads to patterns of peer delinquency trajectories or, alternatively, patterns of peer delinquency trajectories influence late adolescent delinquent behavior. sujung.cho@siu.edu

Insun Park is an assistant professor in the Departments of Criminal Justice and Sociology at the University of Akron. Her research interests include juvenile delinquency, youth in the justice system, theory testing in criminology, and cross-national criminal justice system. Her research interests are how individual internal and external characteristics impact to crime and deviant behavior. ipark@uakron.edu

References

Ashrafi

Feng

C. X.

Neudorf

Alphonsus

K. B.

(2020). Bullying victimization among preadolescents in a community-based sample in Canada: A latent class analysis. BMC Research Notes, 13(1), 1–6. https://doi.org/10.1186/s13104-020-04989-4

Asparouhov

Muthén

(2010). Bayesian analysis using Mplus: Technical implementation (Mplus technical report). http://statmodel.com/download/Bayes3.pdf

Augustine

M. C.

Wilcox

Ousey

G. C.

Clayton

R. R.

(2002). Opportunity theory and adolescent school-based victimization. Violence and Victims, 17(2), 233–253. https://doi.org/10.1891/vivi.17.2.233.33643

Berkowitz

De Pedro

K. T.

Gilreath

T. D.

(2015). A latent class analysis of victimization among middle and high school students in California. Journal of School Violence, 14(3), 316–333. https://doi.org/10.1080/15388220.2014.980424

Borowsky

I. W.

Taliaferro

L. A.

McMorris

B. J.

(2013). Suicidal thinking and behavior among youth involved in verbal and social bullying: Risk and protective factors. Journal of Adolescent Health, 53(1), S4–S12. https://doi.org/10.1016/j.jadohealth.2012.10.280

Bradshaw

C. P.

Waasdorp

T. E.

Johnson

S. L.

(2015). Overlapping verbal, relational, physical, and electronic forms of bullying in adolescence: Influence of school context. Journal of Clinical Child & Adolescent Psychology, 44(3), 494–508. https://doi.org/10.1080/15374416.2014.893516

Bradshaw

C. P.

Waasdorp

T. E.

O’Brennan

L. M.

(2013). A latent class approach to examining forms of peer victimization. Journal of Educational Psychology, 105(3), 839–849. https://doi.org/10.1037/a0032091

Bryk

A. S.

Raudenbush

S. W.

(1992). Hierarchical linear models: Applications and data analysis methods. Sage Publications, Inc.

Butler

L. C.

Kulig

T. C.

Fisher

B. S.

Wilcox

(2019). Victimization at schools and on college and university campuses: Historical developments and applications of the opportunity framework Handbook on crime and deviance (pp. 53–84). Springer. https://doi.org/10.1007/978-3-030-20779-3_4

10.

Cho

(2017a). Explaining the overlap between bullying perpetration and bullying victimization: Assessing the time-ordered and correlative relationships. Children and Youth Services Review, 79, 280–290. https://doi.org/10.1016/j.childyouth.2017.06.016

11.

Cho

(2017b). Self-control and risky lifestyles in context: Cross-level integration between opportunity and collective efficacy in the study of peer victimization among South Korean youth. Journal of Child and Family Studies, 26(1), 67–79. https://doi.org/10.1007/s10826-016-0554-y

12.

Cho

(2018). The impact of low self-control and delinquent peer associations on bullying perpetration and victimization among South Korean adolescents: Time-concurrent, time-lagged, and latent growth curve modeling. Journal of School Violence, 17(4), 500–520. https://doi.org/10.1080/15388220.2018.1453821

13.

Cho

(2019). Self-control, risky lifestyles, and bullying victimization among Korean youth: Estimating a second-order latent growth model. Journal of Child and Family Studies, 28(1), 2131–2144. https://doi.org/10.1007/s10826-019-01432-6

14.

Cho

Lee

J. R.

(2021). Impacts of low self-control and delinquent peer associations on bullying growth trajectories among Korean youth: A latent growth mixture modeling approach. Journal of Interpersonal Violence, 36(7–8), NP4139–NP4169. https://doi.org/10.1177/0886260518786495

15.

Cho

Wooldredge

(2018). Lifestyles, informal controls, and youth victimization risk in South Korea and the United States. Journal of Child and Family Studies, 27(4), 1358–1371. https://doi.org/10.1007/s10826-017-0973-4

16.

Choi

Kruis

Lee

(2020). Applying target congruence theory to victimization risk of students from multicultural backgrounds: A comparison of South Korean, North Korean, and other multicultural family adolescents. Children and Youth Services Review, 118, 105392. https://doi.org/10.1016/j.childyouth.2020.105392

17.

Choi

Kruis

Lee

(2021). Target congruence as a means of understanding the risk of bullying victimization among multicultural family youth in South Korea. Crime & Delinquency. Online ahead of print https://doi.org/10.1177/00111287211022632

18.

Choi

K. S.

Cho

Lee

J. R.

(2019). Impacts of online risky behaviors and cybersecurity management on cyberbullying and traditional bullying victimization among Korean youth: Application of cyber-routine activities theory with latent class analysis. Computers in Human Behavior, 100, 1–10. https://doi.org/10.1016/j.chb.2019.06.007

19.

Cohen

L. E.

Felson

(1979). Social change and crime rate trends: A routine activity approach. American Sociological Review, 44(4), 588–608. https://doi.org/10.2307/2094589

20.

Cohen

L. E.

Kluegel

J. R.

Land

K. C.

(1981). Social inequality and predatory criminal victimization: An exposition and test of a formal theory. American Sociological Review, 46(5), 505–524. https://doi.org/10.2307/2094935

21.

Due

Holstein

B. E.

Lynch

Diderichsen

Gabhain

S. N.

Scheidt

Currie

(2005). Bullying and symptoms among school-aged children: International comparative cross sectional study in 28 countries. European Journal of Public Health, 15(2), 128–132. https://doi.org/10.1093/eurpub/cki105

22.

Ferreira-Junior

Valente

J. Y.

Sanchez

Z. M.

(2021). Examining associations between race, gender, alcohol use, school performance, and patterns of bullying in the school context: A latent class analysis. Journal of Interpersonal Violence, 886260521999123. Online ahead of print https://doi.org/10.1177/0886260521999123

23.

Finkelhor

Asdigian

N. L.

(1996). Risk factors for youth victimization: Beyond a lifestyles/routine activities theory approach. Violence and Victims, 11(1), 3–20. https://doi.org/10.1891/0886-6708.11.1.3

24.

Frías

S. M.

Finkelhor

(2017). Victimizations of Mexican youth (12–17 years old): A 2014 national survey. Child Abuse & Neglect, 67, 86–97. https://doi.org/10.1016/j.chiabu.2017.02.013

25.

Gaffney

Farrington

D. P.

Ttofi

M. M.

(2019). Examining the effectiveness of school-bullying intervention programs globally: A meta-analysis. International Journal of Bullying Prevention, 1(1), 14–31. https://doi.org/10.1007/s42380-019-0007-4

26.

Gladden

R. M.

Vivolo-Kantor

A. M.

Hamburger

M. E.

Lumpkin

C. D.

(2014). Bullying surveillance among youths: Uniform definitions for public health and recommended data elements. Version 1.0. National Center for Injury Prevention and Control, Centers for Disease Control and Prevention and U.S. Department of Education. https://stacks.cdc.gov/view/cdc/21596

27.

Graham

J. W.

(2012). Missing data: Analysis and design. Springer Science & Business Media. https://doi.org/10.1007/978-1-4614-4018-5

28.

Haynie

D. L.

Nansel

Eitel

Crump

A. D.

Saylor

Simons-Morton

(2001). Bullies, victims, and bully/victims: Distinct groups of at-risk youth. The Journal of Early Adolescence, 21(1), 29–49. https://doi.org/10.1177/0272431601021001002

29.

Henson

Wilcox

Reyns

B. W.

Cullen

F. T.

(2010). Gender, adolescent lifestyles, and violent victimization: Implications for routine activity theory. Victims & Offenders, 5(4), 303–328. https://doi.org/10.1080/15564886.2010.509651

30.

Hindelang

M. J.

Gottfredson

M. R.

Garofalo

(1978). Victims of personal crime: An empirical foundation for a theory of personal victimization. Ballinger.

31.

Holt

M. K.

Espelage

D. L.

(2007). Perceived social support among bullies, victims, and bully-victims. Journal of Youth and Adolescence, 36(8), 984–994. https://doi.org/10.1007/s10964-006-9153-3

32.

Hong

J. S.

Kim

D. H.

Piquero

A. R.

(2017). Assessing the links between punitive parenting, peer deviance, social isolation and bullying perpetration and victimization in South Korean adolescents. Child Abuse & Neglect, 73, 63–70. https://doi.org/10.1016/j.chiabu.2017.09.017

33.

B. H.

G. H.

(2003). Development and validation of emotional or behavior problems scale. Korean Journal of Counseling and Psychotherapy, 15(4), 729–746.

34.

Jose

P. E.

Kljakovic

Scheib

Notter

(2012). The joint development of traditional bullying and victimization with cyber bullying and victimization in adolescence. Journal of Research on Adolescence, 22(2), 301–309. https://doi.org/10.1111/j.1532-7795.2011.00764.x

35.

Juvonen

Graham

Schuster

M. A.

(2003). Bullying among young adolescents: The strong, the weak, and the troubled. Pediatrics, 112(6), 1231–1237. http://pediatrics.aappublications.org/content/112/6/1231

36.

Kahle

Peguero

A. A.

(2017). Bodies and bullying: The interaction of gender, race, ethnicity, weight, and inequality with school victimization. Victims & Offenders, 12(2), 323–345. https://doi.org/10.1080/15564886.2015.1117551

37.

Kim

J. Y.

Sung

S. M.

Kim

J. B.

Park

Youn

B. S.

(2015). Bully-victims and their depression, aggression, self-control and experience of domestic violence. Journal of School Social Work, 31, 83–109. https://doi.org/10.1016/j.bonr.2015.10.003

38.

Kim

S. G.

Yun

Kim

J. H.

(2016). Associations between body weight and bullying among South Korean adolescents. The Journal of Early Adolescence, 36(4), 551–574. https://doi.org/10.1177/0272431615577204

39.

Korean Ministry of Education (2019). 2019 National bullying survey. https://moe.go.kr/boardCnts/view.do?boardID=294&boardSeq=78346&lev=0&searchType=null&statusYN=W&page=1&s=moe&m=020402&opType=N

40.

Korean Ministry of Education (2020). Education basic statistics 2020. https://www.moe.go.kr/boardCnts/view.do?boardID=294&boardSeq=81682&lev=0&searchType=null&statusYN=W&page=22&s=moe&m=020402&opType=N

41.

Kulig

T. C.

Pratt

T. C.

Cullen

F. T.

Chouhy

Unnever

J. D.

(2017). Explaining bullying victimization: Assessing the generality of the low self-control/risky lifestyle model. Victims & Offenders, 12(6), 891–912. https://doi.org/10.1080/15564886.2017.1307297

42.

Lasky

N. V.

Peterson

Wilcox

Fisher

B. S.

(2021). Examining patterns of school-based polyvictimization and repeat victimization: A latent class analysis approach. Victims & Offenders, 16(5), 723–745. https://doi.org/10.1080/15564886.2020.1851331

43.

Lee

(2017). A longitudinal study of bullying victimization and depression: Mediating effects of self-esteem and self-resilience. Journal of the Korean Society of School Health, 30(3), 274–283. https://doi.org/10.15434/kssh.2017.30.3.274

44.

Lee

(2014). Psychosocial characteristics of bullies, victims, bully-victims among elementary school students. Korean Youth Research, 21(5), 391–416.

45.

Liu

Guo

Weissman

Liu

(2021). Investigating factors associated with bullying utilizing latent class analysis among adolescents. School Psychology International, 42(1), 11–32. https://doi.org/10.1177/0143034320967061

46.

Lubke

Muthén

B. O.

(2007). Performance of Factor Mixture Models as a Function of Model Size, Covariate Effects, and Class-Specific Parameters. Structural Equation Modeling: A Multidisciplinary Journal, 14(1), 26–47. https://doi.org/10.1080/10705510709336735

47.

Martínez-Monteagudo

M. C.

Delgado

Inglés

C. J.

Escortell

(2020). Cyberbullying and social anxiety: A latent class analysis among Spanish adolescents. International Journal of Environmental Research and Public Health, 17(2), 406. https://doi.org/10.3390/ijerph17020406

48.

McCutcheon

A. L.

(1987). Latent class Analysis. Sage.

49.

Molenaar

P. C.

Campbell

C. G.

(2009). The new person-specific paradigm in psychology. Current Directions in Psychological Science, 18(2), 112–117. https://doi.org/10.1111/j.1467-8721.2009.01619.x

50.

Moon

Alarid

L. F.

(2015). School bullying, low self-control, and opportunity. Journal of Interpersonal Violence, 30(5), 839–856. https://doi.org/10.1177/0886260514536281

51.

Muthén

L. K.

Muthén

B. O.

(2015). Mplus statistical software version 7.31.

52.

Nagin

D. S.

(2005). Group-based modeling of development. Harvard University Press.

53.

Nocentini

Palladino

B. E.

Menesini

(2019). For whom is anti-bullying intervention most effective? The role of temperament. International Journal of Environmental Research and Public Health, 16(3), 388. https://doi.org/10.3390/ijerph16030388

54.

Nylund

Bellmore

Nishina

Graham

(2007b). Subtypes, severity, and structural stability of peer victimization: What does latent class analysis say? Child Development, 78(6), 1706–1722. https://doi.org/10.1111/j.1467-8624.2007.01097.x

55.

Nylund

K. L.

Asparouhov

Muthén

B. O.

(2007a). Deciding on the number of classes in latent class analysis and growth mixture modeling: A Monte Carlo simulation study. Structural Equation Modeling: A Multidisciplinary Journal, 14(4), 535–569. https://doi.org/10.1080/10705510701575396

56.

Olweus

(1993). Bullying at school: What we know and what we can do. Blackwell.

57.

Olweus

(1994). Bullying at school: Basic facts and effects of a school based intervention program. Journal of Child Psychology and Psychiatry, 35(7), 1171–1190. https://doi.org/10.1111/j.1469-7610.1994.tb01229.x

58.

Olweus

(2010). Understanding and researching bullying: Some critical issues. In Jimerson

S. M. S. S. R.

Espelage

D. L.

(Eds.), Handbook of bullying in schools: An International perspective (pp. 9–33). Routledge.

59.

Park

Cho

(2021). Influences of target congruence and lifestyles on victimization trajectories: An autoregressive latent trajectory modeling approach to overlap between bullying and victimization. Journal of Adolescence, 92, 57–74. https://doi.org/10.1016/j.adolescence.2021.08.005

60.

Pellegrini

A. D.

Long

J. D.

(2002). A longitudinal study of bullying, dominance, and victimization during the transition from primary school through secondary school. British Journal of Developmental Psychology, 20(2), 259–280. https://doi.org/10.1348/026151002166442

61.

Pepler

D. J.

Craig

W. M.

Connolly

J. A.

Yuile

McMaster

Jiang

(2006). A developmental perspective on bullying. Aggressive Behavior: Official Journal of the International Society for Research on Aggression, 32(4), 376–384. https://doi.org/10.1002/ab.20136

62.

Radloff

L. S.

(1977). The CES-D scale: A self-report depression scale for research in the general population. Applied Psychological Measurement, 1(3), 385–401. https://doi.org/10.1177/014662167700100306

63.

Savolainen

Brauer

J. R.

Ellonen

(2020). Beauty is in the eye of the offender: Physical attractiveness and adolescent victimization. Journal of Criminal Justice, 66, 101652. https://doi.org/10.1016/j.jcrimjus.2019.101652

64.

Seo

H. J.

Jung

Y. E.

Kim

M. D.

Bahk

W. M.

(2017). Factors associated with bullying victimization among Korean adolescents. Neuropsychiatric Disease and Treatment, 13, 2429–2435. https://doi.org/10.2147/NDT.S140535

65.

Stutzenberger

A. L.

(2020). Exploring pathways of bullying victimization: A test of two competing victimization theories to better understand risk of bullying experiences among middle school youth. [Unpublished doctoral dissertation]. University of Cincinnati.

66.

Tein

J. Y.

Coxe

Cham

(2013). Statistical power to detect the correct number of classes in latent profile analysis. Structural Equation Modeling: A Multidisciplinary Journal, 20(4), 640–657. https://doi.org/10.1080/10705511.2013.824781

67.

Turanovic

J. J.

Pratt

T. C.

(2014). Can’t stop, won’t stop”: Self-control, risky lifestyles, and repeat victimization. Journal of Quantitative Criminology, 30(1), 29–56. https://doi.org/10.1007/s10940-012-9188-4

68.

Y. J.

Choi

Y. J.

Yoo

S. Y.

(2019). Relationships between smartphone dependency and aggression among middle school students: Mediating and moderating effects of ego-resilience, parenting behaviour, and peer attachment. International Journal of Environmental Research and Public Health, 16(19), 3534–3553. https://doi.org/10.3390/ijerph16193534

69.

Volk

Craig

Boyce

King

(2006). Adolescent risk correlates of bullying and different types of victimization. International Journal of Adolescent Medicine and Health, 18(4), 575–586. https://doi.org/10.1515/IJAMH.2006.18.4.575

70.

Waasdorp

T. E.

Bradshaw

C. P.

(2015). The overlap between cyberbullying and traditional bullying. Journal of Adolescent Health, 56(5), 483–488. https://doi.org/10.1016/j.jadohealth.2014.12.002

71.

Wang

Iannotti

R. J.

Luk

J. W.

Nansel

T. R.

(2010). Co-occurrence of victimization from five subtypes of bullying: Physical, verbal, social exclusion, spreading rumors, and cyber. Journal of Pediatric Psychology, 35(10), 1103–1112. https://doi.org/10.1093/jpepsy/jsq048

72.

Widom

C. S.

(1989). Child abuse, neglect, and adult behavior: Research design and findings on criminality, violence, and child abuse. American Journal of Orthopsychiatry, 59(3), 355–367. https://doi.org/10.1111/j.1939-0025.1989.tb01671.x

73.

Yang

S. J.

Kim

J. M.

Kim

S. W.

Shin

I. S.

Yoon

J. S.

(2006). Bullying and victimization behaviors in boys and girls at South Korean primary schools. Journal of the American Academy of Child and Adolescent Psychiatry, 45(1), 69–77. https://doi.org/10.1097/01.chi.0000186401.05465.2c

74.

Zavala

(2020). When the cop is the victim: A test of target congruence theory on intimate Partner violence victimization experienced by police officers. Journal of Interpersonal Violence, 35(17–18), 3622–3642. https://doi.org/10.1177/0886260517709801

75.

Zavala

Whitney

(2019). Target congruence theory and Latino youth victimization. Journal of Ethnicity in Criminal Justice, 17(2), 169–185. https://doi.org/10.1080/15377938.2019.1584137